frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

Open in hackernews

Show HN: Dhansishtha-2.0-preview – First Intermediate Reasoning model

https://helpingai.co
4•VarunGuptaHAI•5h ago
Hey HN,

We’re the team behind HelpingAI, and we’ve been experimenting with ways to make LLM reasoning faster and more efficient, especially for complex tasks like math problems, coding questions, and logic puzzles.

During a fine-tuning session of our older reasoning model, a bug accidentally introduced a think tag mid-response. Rather than break the model, it created a chain-of-thought style intermediate step — and the model got better. That bug inspired us to explore what we now call Intermediate Reasoning.

We scaled this up by fine-tuning Qwen3-14B on 3 trillion tokens of reasoning-specific data — curated for multistage reasoning tasks — and the results blew us away:

• 5x faster reasoning vs. models like DeepSeek-R1, Grok, and OpenChat.

• Significantly lower token usage and latency, ideal for startups and real-time use cases.

• Solves benchmark-level math and logic problems in seconds instead of minutes.

On a hard math benchmark, for example:

• Dhanishtha-2.0 solved it in 45s

• DeepSeek-R1 took 280+ seconds

You can try it here:

• Chat/API access: https://helpingai.co

• Model weights (open source): https://huggingface.co/HelpingAI/Dhanishtha-2.0-preview

We’d love feedback, ideas, and especially edge cases where you think it might break (or surprise you). Happy to answer any questions and go into training data, fine-tuning strategy, evals, or infra!

Thanks,

Varun Gupta

Co-founder, HelpingAI

Effective TikTok Content Strategies for Entrepreneurs

1•emmanol•58s ago•0 comments

The Cangjie Programming Language by Huawei

https://cangjie-lang.cn/en/docs?url=%2F0.53.13%2Fwhite_paper%2Fsource_en%2Fcj-wp-abstract.html
1•thunderbong•3m ago•0 comments

Can Large Language Models Play Text Games Well?

https://arxiv.org/abs/2304.02868
2•willvarfar•6m ago•0 comments

China's giant new gamble with digital IDs

https://www.economist.com/china/2025/07/01/chinas-giant-new-gamble-with-digital-ids
1•bookofjoe•6m ago•1 comments

Trump's big bill will change taxes

https://thehill.com/business/5381924-trump-tax-cuts-bill-breakdown/
1•koolba•6m ago•1 comments

Show HN: I'll build your billing stack in 24 hours – ping me

1•vijay_gorfad•12m ago•0 comments

Designing a Life Management System That Doesn't Fight Back

https://medium.com/@chrisveleris/designing-a-life-management-system-that-doesnt-fight-back-2fd58773e857
1•cvicpp123•14m ago•0 comments

Brine Rejection

https://en.wikipedia.org/wiki/Brine_rejection
3•Matumio•15m ago•0 comments

Show HN: We built an AI-agent with a state machine instead of a giant prompt

https://nomos.dowhile.dev/
1•chandralegend•15m ago•0 comments

The masterful design of the two-liter plastic soda bottle [video]

https://www.youtube.com/watch?v=kU_gH36GG58
1•luismedel•17m ago•0 comments

With all the cuts to NASA and STEM fields,are we giving up on Tech race to China

2•FrustratedMonky•24m ago•1 comments

AI That Thinks Like Us – and Could Help Explain How We Think

https://www.helmholtz-munich.de/en/newsroom/news-all/artikel/ai-that-thinks-like-us-and-could-help-explain-how-we-think
1•doener•24m ago•0 comments

2k Verified CEO Emails from $5M+-Revenue Companies (Free)

https://laiskas.com/examples
1•Simonekis•25m ago•0 comments

Show HN: A tool to simplify invoicing in Stripe (invoicelyapp.com)

https://invoicelyapp.com/
1•creativehubspac•26m ago•0 comments

The chemical secrets that help keep honey fresh for so long

https://www.bbc.com/future/article/20250701-the-chemical-secrets-that-help-keep-honey-fresh-for-so-long
2•bookofjoe•26m ago•0 comments

Your Brain on AI

https://aibrain.joaquinbernal.com
1•joaquinbernal•26m ago•0 comments

Microsoft: "ChatGPT isn't better than Copilot,"

https://www.windowscentral.com/software-apps/chatgpt-more-visits-in-a-day-than-copilot-in-a-month
3•Bluestein•29m ago•0 comments

Show HN: Refractive JavaScript Glass

https://github.com/Explosion-Scratch/realglass
1•explosion-s•34m ago•0 comments

Show HN: Nich digital catalogs and mockups that make pattern designers money

https://patternisse.com
1•ferreirarmp•37m ago•0 comments

OpenAI almost shipped ChatGPT with a different name – before a late-night twist

https://www.windowscentral.com/artificial-intelligence/openai-chatgpt/chatgpt-almost-shipped-with-a-different-name
4•Bluestein•39m ago•0 comments

Reflections on Haskell and Rust

https://academy.fpblock.com/blog/rust-haskell-reflections/
1•runeks•39m ago•1 comments

Show HN: Rhythm Is All You Need: Someone Please Build This

https://medium.com/@rikudatian0/why-music-moves-you-and-ai-cant-understand-it-dbccdbe33c23
2•RyukuLogos•39m ago•2 comments

Your Review: Alpha School

https://www.astralcodexten.com/p/your-review-alpha-school
3•Anon84•39m ago•0 comments

New World Bank classification of countries by income groups

https://blogs.worldbank.org/en/opendata/understanding-country-income--world-bank-group-income-classifica
4•alphabetatango•43m ago•1 comments

Establishing Best Practices for Building Rigorous Agentic Benchmarks

https://arxiv.org/abs/2507.02825
1•consumer451•44m ago•0 comments

Microsoft to Lay Off About 9k Employees

https://www.nytimes.com/2025/07/02/technology/microsoft-layoffs-ai.html
3•mitchbob•49m ago•1 comments

India's Scramble for AI Independence

https://www.technologyreview.com/2025/07/04/1119705/inside-indias-scramble-for-ai-independence/
1•Brajeshwar•49m ago•0 comments

Home – AniWave – Watch Free Anime, Online Anime Streaming – Aniwave

https://www.aniwave.se/
1•nkush•49m ago•0 comments

Cuba's Power Grid Nears Total Failure

https://spectrum.ieee.org/cuba-energy-crisis
1•Brajeshwar•50m ago•0 comments

Unify AI Coding Tools with One Standard

1•level09•50m ago•0 comments